ESEM observation and rheological analysis of rejuvenated SBS modified bitumen

(1)

ESEM observation and rheological analysis of rejuvenated SBS modified bitumen

Lin, P.; Liu, X.; Apostolidis, P.; Erkens, S.; Zhang, Y. ; Ren, S.

DOI

10.1016/j.matdes.2021.109639

Publication date

2021

Document Version

Final published version

Published in

Materials & Design

Citation (APA)

Lin, P., Liu, X., Apostolidis, P., Erkens, S., Zhang, Y., & Ren, S. (2021). ESEM observation and rheological

analysis of rejuvenated SBS modified bitumen. Materials & Design, 204, 1-18. [109639].

https://doi.org/10.1016/j.matdes.2021.109639

Important note

To cite this publication, please use the final published version (if applicable).

Please check the document version above.

Copyright

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons. Takedown policy

Please contact us and provide details if you believe this document breaches copyrights. We will remove access to the work immediately and investigate your claim.

This work is downloaded from Delft University of Technology.

(2)

ESEM observation and rheological analysis of rejuvenated SBS

modi

ﬁed bitumen

P. Lin

a,

⁎

,

X. Liu

a

, P. Apostolidis

a

, S. Erkens

a

, Y. Zhang

b

, S. Ren

a a

Civil Engineering and Geosciences, Delft University of Technology, Delft 2628CN, the Netherlands

b_{School of Highway, Chang'an University, Xi'an 710064, Shaanxi, China}

H I G H L I G H T S

• The addition of rejuvenators contributes to the formation offibril structure and increases the distance betweenfibrils. • The influence of rejuvenator

types/dos-ages on theﬁbril-structure can be dem-onstrated with ESEM.

• A process for extracting quantiﬁcation parameters from theﬁbril-structure of bitumen is proposed.

• The ﬁbril-structure parameters exhib-ited a signiﬁcant correlation with rheo-logical properties. G R A P H I C A L A B S T R A C T

a b s t r a c t

a r t i c l e i n f o

Article history: Received 10 September 2020 Received in revised form 4 March 2021 Accepted 5 March 2021

Available online 08 March 2021

Observing the microstructure of bituminous binders with an environmental scanning electron microscope (ESEM) can contribute significantly to reveal the underlying rejuvenation mechanism. In this study, three rejuvenators were selected to regenerate the aged SBS modified binders at five dosages, and their rheology was evaluated using a dynamic shear rheometer. ESEM was employed to examine the microstructure of binders as well, and a se-ries of microstructure parameters were quantifiedwith image analysis. Theresults demonstratedthatthechemical composition changes correspond to the evolution of microstructure morphological and rheological properties. Moreover, the rheological and microstructure characteristics were analyzed with Principal Component Analysis (PCA) and regression analysis. Based on PCA results, the microstructure of rejuvenated binders has shown a good correlation with stiffness after combining various principal components. According to regression analysis, the dis-tance between adjacentfibrils exhibited a significant correlation with Jnr3.2and the complex modulus index.

Over-all, the results of this study strengthen the hypothesis that the ESEM microstructure is intimately correlated with chemical composition and rheological properties, rather than with irrelevant surface phenomena.

Bitumen, polymer modiﬁed bitumen, rejuvenator, microstructure, rheology, environmental scanning electron microscopy.

1. Introduction

Asphalt binder is the main construction material forﬂexible pave-ments, and the worldwide production of asphalt materials is estimated

to be more than a billion tons in 2019 [1,2]. End-of-service-life asphalt pavement is milled and collected, producing a large amount of reclaimed asphalt pavement (RAP). RAP application in new pavements can conserve valuable natural resources and reduce greenhouse gas emissions [3,4].

During aging, the elemental ratios and chemical components of the bitumen on the RAP are changed, which results in the deterioration of

⁎ Corresponding author.

E-mail address:P.lin-2@tudelft.nl(P. Lin).

https://doi.org/10.1016/j.matdes.2021.109639

Contents lists available atScienceDirect

Materials and Design

(3)

the physical and rheological properties of bitumen [5,6]. In the recycling of RAP, the addition of the rejuvenator is an important way to recover the physical and rheological properties of aged binders [7_–11]. There are plenty of studies about the effect of rejuvenators on binder rheology. Rejuvenators soften the aged bitumen causing the increase of penetra-tion, ductility, phase angle and colloidal instability index, and decrease of viscosity and complex modulus [3,4,12–18]. Meanwhile, recent in-vestigations have shown a close relationship of microstructure with chemical composition and rheological properties of bitumen [19–23], assisting on understanding the rejuvenation mechanism. However, the rejuvenator's effect on the microstructure of aged bitumen is not well understood yet.

The bitumen microstructure has been intensively investigated with different microscopic methods, such as optical microscopy [24,25], fluo-rescence microscopy [26–28], transmission electron microscopy [28], atomic force microscopy [29_–31] and scanning electron microscopy (SEM) [32–34]. Among these methods, SEM is a promising one to inves-tigate the microstructure morphologies of various binders [35,36]. In the 1980s, the presence of environmental scanning electron microscope (ESEM) improved the observation quality for poorly conductive and vis-cous oil-bearing specimens, such as bitumen [37–39]. The advantage of the ESEM in the research for oil-bearing materials such as bitumen is that deposition of a conductive metal alloy or carbon coating onto the sample is not required, and the surfaces can be imaged directly. This al-lows one to conduct in situ studies of the electron-sample and gas-sample interactions as well as fracture studies, where new surfaces are generated during crack propagation [40]. During the examination of bitumenfilms with ESEM, a modification of the bitumen surface was observed after exposingfilms to the electron beam for several mi-nutes. The bitumen surface is initiallyflat but appears as a random net-work of fibrils after beam exposure [41]. It is envisioned that the electron beam preferentially etches the low molecular weight oils (sat-urate and aromatic) from bitumen and the remaining portion consists of the higher molecular weight components, usually the asphaltenes and resins. Due to this reason, thefibril structure is highly influenced by the fractions of saturates, aromatics, resins and asphaltenes (SARA) of bitumen, described with colloidal instability index [27].

In the 1990s, a series of investigations reported that“random net-works offibrils” can be observed after several minutes of electron beam exposure with secondary electron signal mode [40–42]. Rozeveld et al. explained this phenomenon as the light molecular component's volatilization due to the localized heating generated by electron beam (200 °C) [41]. The“networks of fibrils” were assumed to be a mix of asphaltenes and resins. Thefibril structure was changed by tensile load-ing, with the strands in the structure aligning themselves parallel to the tensile direction. Stangl et al. investigated the microstructure of bitu-men with ESEM and nano-indention tests and found the distance be-tween two adjacent strands in these two different tests was the same, indicating that thefibril structure was not induced by the electron beam [43]. Also, others agreed that the ESEM morphology shows the structure of bitumen rather than the electron beam caused artifact [41,42,44]. Gaskin et al. reported that the electron beam could activate the radiolytic particles and evaporate the light components from the material surface [45].Gaskin also reported the ESEM microstructure of bitumen's SARA fractions and waxes. The maltene fraction exhibited a fibrils structure ubiquitous, which was similar to the structure in bitu-men. The maltene might not be free from asphaltene due to the limita-tion of the extraclimita-tion technique; at least, thefibril structure also contained the amount of maltene [45]. Recently, Lu characterized the surface of the structured area with time-of-flight secondary ion mass spectrometry (TOF-SIMS), which demonstrated that the aliphatic and aromatic content is higher and lower, respectively, on the surface of the structured area compared with those of unstructured surfaces. With the Principal Component Analysis (PCA), it was speculated that the worm structure might be attributed to the evaporation of volatiles, hardening and local contraction, and possibly to chemical reactions

(e.g., breaking of chemical bonds, chain scission and crosslinking). In summary, the formation of theﬁbril structure and its composition is still not totally clear.

According to other ESEM studies, the bitumen microstructure is also influenced by aging and modification. Rozeveld et al. reported that the addition of styrene-butadiene rubber in bitumen could lead to a signif-icant change of microstructure from a random pattern to textured. Meanwhile, the aging of base bitumen led to a coarserfibril structure and a largerfibril diameter [41]. Mikhailenko et al. reported that with the increased oxidation degree, thefibril structure was denser, more structured and smaller infibril diameter [46]. The different conclusions in these two studies may be due to the different origin of base bitumens, indicating the influence of aging on the microstructure of bitumens needs further investigation. The addition of styrene-butadiene-styrene increased thefibril diameter and microstructure density [47]. Similar observations were obtained elsewhere [34,43,44,48]. Stangl found that the packing density of microstructure obtained by ESEM correlates well with the change in gel permeation chromatography (GPC) results and viscosity increase [43]. Mikhailenko also reported that the_“fibril area” and “formation time” had a good correlation with penetration value and softening point [47]. Overall, the evolution of physical proper-ties corresponds to bitumen microstructure, and the latter can be used as the“fingerprint” of bituminous binders.

Although there are interestingfindings in ESEM, limited studies exist on assessing rejuvenators' influence on bitumen. In this study, particular emphasis will be given to providing insights into the rejuvenation mechanism of base and polymer modified bituminous binders, collec-tively called binders, using ESEM. Particularly, the effect of aging and re-juvenation on the binder microstructure will be evaluated. Moreover, evidence will be provided on rejuvenators' influence on binder micro-structure. Finally, a correlation between rheology and ESEM microstruc-tural characteristics will be proposed.

2. Materials and methods

2.1. Binders, rejuvenators and aging protocols

The styrene-butadiene-styrene (SBS) is one of the most widely used thermoplastic polymers for bitumen modiﬁcation, which can improve the material rutting and thermal cracking resistance [49–51]. A com-mercial SBS polymer modiﬁed bitumen (PMB) was used in this re-search. The basic rheological and physical properties of PMB are given

inTable 1.

Three different products were selected to investigate the effect of re-juvenators on the rheological properties and ESEM microstructural characteristics of aged PMB. The rejuvenator Rej-A is an industrial reju-venator rich in aromatic compounds. According to the material supplier,

Table 1

Basic rheological and physical properties of PMB binders.

Properties Measurement Technical

Criterion Speciﬁcations Original Penetration at 25 °C, 100 g, 5 s (0.1-mm) 55 40– 60 ASTM-D5

Softening point (°C) 65.7 >60 ASTM-D36 G*/sinδ @ 70 °C (kPa) 2.408 >1.0 kPa ASTM-D6373 G*/sinδ @ 76 °C (kPa) 1.419 >1.0 kPa ASTM-D6373 Glass transition temperature (°C) −15.4 – –

RTFOT

Mass loss (%) 0.22 ± 1.0

G*/sinδ @ 70 °C (kPa) 2.905 >2.2 kPa ASTM-D6373 G*/sinδ @ 76 °C (kPa) 1.607 >2.2 kPa ASTM-D6373 Glass transition temperature (°C) −14.8 – –

(4)

the aromatic type and ratio in Rej-A are designed based on the Hansen solubility parameter, and Rej-A is especially suited for high RAP mate-rials produced at normal asphalt production temperatures (160 °C). Rej-B is a low viscous liquid rejuvenator, rich in saturates and contains selected components for the re-compatibilization of the various phases in oxidized mastic.

Rej-C was fabricated in the lab to rejuvenate the aged PMB. The Rej-C contains saturates, aromatics, SBS copolymer and additives to allow the dispersion of agglomerated asphaltenes. The addition of the saturate component (bio-oil) functions to decrease the viscosity, and the aro-matic component (aroaro-matic oil) helps to recover the viscoelastic char-acteristics of aged PMB. A selected petroleum plasticizer was also added as a stabilizer to improve the miscibility between the polymer and saturated/aromatic oil. Afterward, a plant-extracted oil with cyclic monoterpene was added to disperse the agglomerated asphaltenes. The ratios of different components have been optimized through exten-sive experiments. The properties and description of all three rejuvena-tors can be seen inTable 2.

The binders were_{ﬁrstly subjected to short-term aging with Rolling} Thin Film Oven Test (RTFOT) (EN12607) at 163 °C for 75 min, and sub-sequently, long-term aging was conducted in a Pressure Aging Vessel (PAV) (EN14769) at 100 °C for 80 h. The reason for choosing such a long time is that 80 h PAV aging has shown an equivalent aging degree with a porous asphalt surfacing material after ten years of service [52]. Afterward, the rejuvenated bitumen was prepared by blending the aged bitumen and rejuvenator with a mixer under 170 °C for 10 min. The choice of a relatively wide range of rejuvenator dosage (5– 30%) was based on the possibility to observe the microstructure of PMB with ESEM and the rejuvenation propensity of binders. The description of binders can be seen inTable 3.

2.2. Frequency sweep tests

A dynamic shear rheometer (DSR) was applied to characterize the rheology of binders with oscillatory sinusoidal loading. In this study, the DSR tests were conducted in a parallel plate testing conﬁguration with an 8-mm diameter spindle between 0 and 30 °C and a 25-mm di-ameter spindle between 40 and 80 °C, respectively. Frequency sweep tests from 0.01 to 10 Hz were conducted at 0, 20, 30, 40, 60 and 80 °C with strain amplitude within the linear viscoelastic range. Three repli-cates per material were tested.

The complex modulus master curves were constructed with the Sig-moidal Model [53,54]. The phase angle master curves were constructed with Kramers-Kronig relations, which wasﬁrstly suggested by Booij and Thoone [55,56].

log∣G⁎∣ ¼ ν þ₁ α

þ eðβþγlogωÞ ð1Þ

δ ¼ 90 dlogG_dlog_ω⁎¼ −90 αγ eðβþγlogωÞ

1_{þ e}ðβþγlogωÞ

2 ð2Þ

whereω is the reduced frequency at the reference temperature (rad/ s), |G∗| is the complex modulus (Pa), δ is phase angle (degree), ν is the lower asymptote,α is the difference between the values of the upper and lower asymptote,β and γ deﬁne the shape between the asymptotes

and the location of the inﬂection point (inﬂection point obtained from 10(β/γ)).

Cavalli proposed an aging index, which was defined to incorporate the changes over a large frequency range by calculating the difference of area between the complex modulus master curves of fresh and aged bitumen over a defined frequency range [57]. Based on this idea, the indices of complex modulus (IM) and phase angle (IP) are defined as the ratio of the integral area of master curves of the original binder to that of the rejuvenated one. These two parameters can be applied to characterize the rejuvenation degree. The definition of IMand IPare as follows: AM¼ Z4 −5 log G⁎ð Þdωω ð3Þ AP¼ Z4 −5δ ωð Þdω ð4Þ IM¼ AM−Sample AM−OB 100% ð5Þ IP¼ AP−sample AP−OB 100% ð6Þ

where AMis the integral area of complex modulus, AM-OBand AM-sample are the AMof original binder (OB) and rejuvenated binders, APis the in-tegral area of phase angle, AP-OBand AP-sampleare the APof original binder (OB) and rejuvenated binder.

The scheme of the deﬁnition of complex modulus and phase angle in-dices can be seen inFig. 1. When the complex modulus index reaches 100%, the rejuvenated binder will be similar to the original binder accord-ing to the complex modulus master curves. These indices provide methods for rejuvenator selection andrejuvenator dosage determination. 2.3. Multiple stress creep recovery tests

The multiple stress creep recovery (MSCR) tests were conducted to evaluate binders' rutting resistance with a 25-mm spindle and 1-mm gap. The binder is subjected to an alternate cycle of 1 s of creep and 9 s of recovery, at 0.1 and 3.2 kPa stress levels, respectively. The MSCR tests were conducted 30 cycles, and the high-temperature performance of binders was evaluated with non-recovery compliance (Jnr) and

Table 2

Name and properties of rejuvenators.

Rejuvenator Viscosity at 20 °C (Pa·s) Density (kg/m3

)

Petroleum/Organic Main component Auxiliary components Polarity

Rej-A 0.817 0.955 Petroleum Aromatic None High

Rej-B 0.115 0.928 Organic Aliphatic None None

Rej-C 0.752 0.943 Mixed Aromatic & Aliphatic Polymer & Additves* Low

Note* The function of additive is to separate the agglomeration of asphaltene.

Table 3

Description of studied binders.

Name Composition

Base bitumen Base bitumen before the addition of SBS polymer OB Original SBS polymer modified bitumen RTFOT SBS polymer modified bitumen after RTFOT aging 4PAV SBS polymer modified bitumen after RTFOT aging

and 80 h of PAV aging A5,10,15,20,30 4PAV + 5,10,15,20,30% wt Rej-A B5,10,15,20,30 4PAV + 5,10,15,20,30% wt Rej-B C5,10,15,20,30 4PAV + 5,10,15,20,30% wt Rej-C

(5)

percent recovery (R), whose determination is described in detail in ASTM D7405–08. Three replicates per material were tested.

2.4. Linear amplitude sweep tests

The linear amplitude sweep (LAS) is an accelerated test performed at an intermediate temperature [58,59]. Here, the LAS tests were con-ducted in DSR with the 8-mm parallel plates to evaluate the fatigue property of binders at 20 °C. According to AASHTO TP 101, the LAS test consists of a non-destructive frequency sweep test and a strain am-plitude sweep test. The frequency sweep was conducted with a fre-quency range from 0.2 to 30 Hz at a strain amplitude of 0.1% to determine the undamaged binder properties. The amplitude sweep is

performed with a strain level from 0.1 to 30%. Three replicates per ma-terial were tested.

2.5. Environmental scanning Electron microscopy

For the ESEM evaluation of binders, an 8-mm cylindrical holder was prepared. The binders were heated at 150 °C for 30 min. Subsequently, 0.1 g of material was poured on the sample holder with the help of a spatula (Fig. 2(a)). Then, the bitumen-ﬁlled holders were placed on a hotplate at 170 for about 60 s forﬂattening. Before ESEM analyses, all samples were stored in a cooler at 10 °C overnight.

The microstructure assessment of binders was conducted at 25 °C with an ESEM device (Philips XL 30 ESEM). Samples were placed on

Fig. 1. The scheme of the deﬁnition of complex modulus and phase angle indices.

(6)

the ESEM sample stage, which is under the ESEM detector and electron gun (Fig. 2(b,c)). The microstructure parameters were an acceleration voltage of 20 keV, a spot size of 3.5, a chamber pressure of 1.0 Torr of va-pour and three different magniﬁcation (250×, 500× and 1000×) in sec-ondary electron (SE) mode [34,48]. The sample was exposed to the electron beam for 20 min for tracking the microstructure changes in binders. ESEM observations for each binder sample were performed three times.

3. Results

3.1. Viscoelastic properties analysis

The viscoelastic properties of the original, aged and rejuvenated binders were characterized with master curves of complex modulus and phase angle. Speciﬁcally, as illustrated inFig. 3(a), the complex modulus of RTFOT aged sample increased slightly at high frequency and increased substantially at low frequency compared with the origi-nal sample. The short-term aging mainly led to a modulus increase of PMB at the high frequency (high temperature) range. In terms of phase angle master curves, the original binder exhibited a plateau

zone with log frequencies between−3.5 and − 0.5. According to the re-search of Lu and Airey [60,61], the phase angle plateau zone is caused by the polymer network in PMB. Although the presence of SBS copolymer is more apparent before aging according to the phase angle results in the low-frequency spectrum, the phase angle plateau disappeared after aging, indicating the aging induced degradation of the polymer network [49,62].

As illustrated inFig. 3(b-d), the addition of the rejuvenator led to a signiﬁcant decrease of complex modulus and an increase of phase angle both at high and low frequency. The 4PAV aged sample recovered its viscous behavior with the rejuvenator addition. Although all three re-juvenators were able to decrease the complex modulus and increase the phase angle, their rejuvenation effect was different. Rej-B substantially decreased the complex modulus, and Rej-C signiﬁcantly increased the phase angle at relatively low rejuvenator content.

For further quantitative analysis of the rejuvenator's impact, the IM and IPvalues are calculated and presented inFig. 4. During aging, the 4PAV aged binder showed a 28% increase in IMand 28% increase in IP comparing to the original binder. When rejuvenator content is low, there is no signiﬁcant difference between rejuvenators. However, when rejuvenator content is higher than 15%, Rej-B was more effective

-5 -4 -3 -2 -1 0 1 2 3 4 5 100 101 102 103 104 105 106 107 108 109 Complex Modulus,Pa Log Frequency,Log(Hz) Origin 4PAV A5 A10 A15 A20 A30

(b)

20 30 40 50 60 70 80 90 Phase Angle, -5 -4 -3 -2 -1 0 1 2 3 4 5 100 101 102 103 104 105 106 107 108 109

Complex Modulus,Pa

Log Frequency,Log(Hz)

Origin 4PAV B5 B10 B15 B20 B30

(c)

20 30 40 50 60 70 80 90

Phase Angle,

-5 -4 -3 -2 -1 0 1 2 3 4 5 100 101 102 103 104 105 106 107 108 109 Complex Modulus,Pa Log Frequency,Log(Hz) Origin RTFOT 4PAV 20 30 40 50 60 70 80 90 Phase Angle,

Phase angle plateau zone

(a)

-5 -4 -3 -2 -1 0 1 2 3 4 5 100 101 102 103 104 105 106 107 108 109 Complex Modulus,Pa

Log Frequency,Log(Hz)

Origin 4PAV C5 C10 C15 C20 C30 20 30 40 50 60 70 80 90 Phase Angle,

(d)

Fig. 3. Complex modulus and phase angle master curve of origin, aged and rejuvenated PMB binders at a reference temperature of 30 °C, (a) origin and aged PMB, (b) rejuvenated bitumen with Rej-A; (c) rejuvenated bitumen with Rej\\B; (d) rejuvenated bitumen with Rej\\C.

(7)

in decreasing complex modulus compared with Rej-A, and Rej-C and Rej-A are more efﬁcient in increasing phase angle compared with Rej\\B.

3.2. MSCR analysis

To evaluate the impact of rejuvenator on rutting resistance, the MSCR tests were conducted at 70 °C. The average non-recoverable creep com-pliance (Jnr3.2) and average percent recovery at 70 °C were demonstrated

inFig. 5. The RTFOT aged binder has shown slightly lower R3.2and Jnr3.2

compared with the original binder. However, after 4PAV aging, the R3.2 in-creased, and theJnr3.2decreased dramatically. According to previous stud-ies [52,62], the degradation of the polymer network plays a dominant role in short-term aging at a relatively high temperature (163 °C), causing a decrease in elastic recovery ability. After 80 h of PAV aging, the severe ox-idation of thebitumen phase plays thedominantrole, which increases the stiffness and elasticity of PMB.

With the increase of rejuvenator content, R3.2decreased, and Jnr3.2 in-creased extensively, indicating loss of elastic recovery ability and de-crease of rutting resistance. Among these rejuvenators, the rejuvenated bitumen with Rej-C showed higher R3.2, especially when the rejuvenator content was less than 15%. It might be caused by the addition of SBS copol-ymer in Rej\\C, which could supplement the degraded polymer. How-ever, the rejuvenated binder with Rej-A has shown relatively lower Jnr3.2, indicating that this rejuvenator probably had a less adverse impact on rutting resistance. Most rejuvenated binders satisﬁed the criteria of maximum Jnr3.2as given by AASHTO MP19 for heavy trafﬁc conditions,

except the rejuvenated bitumen with 30% of Rej_{\\B. Although the} addi-tion of the rejuvenator decreases elastic recovery and rutting resistance, it is restoring the original behavior of the bitumen, which tends to be stiffer after aging.

3.3. LAS analysis

The LAS tests were conducted to evaluate the fatigue life of all binders. The fatigue life at 2.5% (Nf2.5) and 5% (Nf5) strain levels are illus-trated inFig. 6. After aging, the Nf5%of PMB decreased from 340 cycles (original binder) to 268 cycles (RTOFT aged binder), andfinally to 143 cycles (4PAV aged binder). The addition of the rejuvenator was ef-fective in recovering the fatigue property of the aged PMB binder, espe-cially when the rejuvenator dosage was more than 10%, the fatigue life of the rejuvenated binder was higher than that of the original binder. Rej-A showed relatively less improvement in Nf5%in these three rejuve-nators,indicating the aromatic component was less efficient in improv-ing fatigue properties. In contrast, Rej-C was more efficient in improving the fatigue properties of the aged binder, probably due to the positive effect of the addition of SBS copolymer. The evolution of Nf2.5%for reju-venated bitumen was consistent with the evolution of Nf5%.

3.4. ESEM analysis

3.4.1. Microstructure morphology

ESEM is a robust imaging technique to characterize the microstructure morphology of various materials, including wet, oily and non-conductive

OB

RTFOT4PAV A5 A10 A15 A20 A30 B5 B10 B15 B20 B30 C5 C10 C15 C20 C30

70 80 90 100 110 120 130 140 100 119 128 121 114 105 99 90 116 110 91 92 77 113 110 101 91 82 100 80 72 75 82 92 93 97 77 82 85 87 91 84 85 88 92 99

I

M

（

%

）

（a）

（b）

OB

70 80 90 100 110

I

P

（

%

）

(8)

substances in their original state [34]. Firstly, three types of rejuvenators were investigated with this technique, and the ESEM images are illustrated

inFig. 7. Initially, there was a“needle” structure on the surface of Rej\\B,

which is similar to the ESEM image of base bitumen incorporating a bee-like structure [34,45]. It was speculated that the bee structures were caused by waxes, which may form crystal networks attributed to diffusion or spontaneous alignment [63]. In this research, Rej-B may contain parafﬁn waxes associated with the“needle”likestructures.However,after3minof illumination, these structures totally disappeared (Fig. 7(b)). Also, no changes were observed when the Rej-B was subjected to 20 min of illumi-nation. Similarly, the Rej-A and Rej-C were subjected to 20 min of illumina-tion, and no obvious structure was observed (Fig. 7(c) and (d)).

Here, after a period of illumination, the binders were investigated with ESEM and aﬁbril structure was observed, as shown inFig. 8. The ﬁbril structures of binders were quite similar as demonstrated in

Fig. 8(a) and (b), respectively. Theﬁbril structure of PMBwas slightly

denser (covers more area of the surface) of larger strand diameter than of base bitumen. The RTFOT aging induced a significant evolution offibril structure in binder, in which the strand diameter became smaller, and the surface became rougher. After 4PAV aging, thefibril structure was not observed even after 20 min of illumination, and a few bubble-like spots appeared (Fig. 8(d)). It is because the electron beam preferentially etches the low molecular weight oils from bitumen

and the remaining portion consists of the higher molecular weight com-ponents, usually the asphaltenes and resins [43,64]. During long-term aging, the content of the evaporable component decreased, and the ﬁ-bril structure cannot be seen anymore. The formation mechanism of the bubble-like spot needs further investigation.

To reveal the influence of rejuvenator on the microstructure of the aged binder, sixteen different rejuvenated binders were investigated in ESEM, and the results are shown inFig. 9. For the rejuvenated bitu-men with Rej-A, the ESEM images wereflat and featureless with less than 15% of rejuvenator, just like the 4PAV aged bitumen. Thefibril structure appeared again when the content of Rej-A was more than 20%. Thefibril structure of A20 (aged PMB binder rejuvenated with 20% of Rej-A) and A30 were rougher and more perpendicular structur-ally than of the original microstructure. 15% of Rej-B was needed in bi-tumen for the re-appearance of the_{fibril structure. The fibril structure} of B15 appears to be“random”, which is similar to the structure of the original PMB. With the increase of Rej-B content, thefibril structure be-came sparse and perpendicular. The addition of 10% of Rej-C could en-able the fibril structure to reappear after the electron beam illumination. This phenomenon agrees with the results of IPand IMin which the addition of 10% Rej-C already had a significant rejuvenation effect. This may be due to the role of additives in Rej-C that dissolve and separate the asphaltene agglomeration. With the increase of Rej-C

40.5

35.0

79.7

56.6

37.3 5.5 5.4

0.0

57.5

32.0 2.5 2.9

0.0

68.6

42.0

19.6

3.0

0.0

0.030 0.020 0.002 0.010 0.022 0.196 0.239 0.813 0.011 0.028 0.3680.436 2.372 0.009 0.031 0.124 0.399 1.234 OB

0 10 20 30 40 50 60 70 80 90

R

3.2

（

%

）

（a）

OB

0.001 0.01 0.1 1 10

J

2. 3r n

a

P

k(

-1

)

（b）

J

nr3.2

=2.0kPa

-1

(9)

content, the strand diameter appeared to be larger, and theﬁbril struc-ture became smoother, sparser and more perpendicular.

3.4.2. Fibril formation time

The ESEM image of bitumen isflat and featureless at the beginning. With the increase of the illumination time, thefibril structure appeared. From the literature, the time period to obtain a stable ESEM image is called formation time [47]. The formation time essentially reflects the possibility and rate of volatilization of the light molecular component on the surface of bitumen, and it correlates well with aging degree [34,48,65]. Hence, formation time has the potential to be a parameter for characterizing the degree of aging and rejuvenation of bitumen. The evolution of ESEM images of rejuvenated PMB was demonstrated

inFig. 10. All the samples were featureless at the beginning, and after

1 min of illumination of the electron beam, B20 showed a blurﬁbril structure. A20 and C20 dimly revealed a little of the outline. When the illumination time was 3 min, theﬁbril structure was revealed and remained unchanged afterward.

To quantitatively analyze the formation time, the evolution of the ESEM image was recorded as a video to precisely extract the formation time data. From previous literature, theﬁbril structure can only be re-vealed when the electron beam etches the evaporable component in the surface layer [41,60]. According to previous research [34], it may be because the lighter components of the 4PAV aged binder were not enough to form the ﬁbril structure. The addition of rejuvenators

supplements the saturate and aromatic components, allowing theﬁbril structure formation also decreasing the formation time. Thus, the for-mation time can be used as a parameter to characterize the degree of re-juvenation. The formation time will not be recorded if the ﬁbril structure did not appear even after 20 min of illumination. As shown

inFig. 11, the formation time increased after RTFOT aging, which is

co-herent with Mikhailenko's_{finding [}65]. At the same time, the addition of rejuvenators can significantly decrease the formation time. Among the three rejuvenators, Rej-B was most effective in decreasing the for-mation time. 30% of Rej-B reduced the forfor-mation time to 42 s, which is much lower than that of the original sample. However, the formation time of A30 was 138 s, more than three times that of B30. The efficiency of Rej-C in decreasing formation time is between Rej-A and Rej\\B. 3.4.3. ESEM image quantification

For further quantitative analysis of the ESEM images, a process was conducted to extract the microscopic parameters offibril structure, which is based on Stangl and Mikhailenko's work [43,48]. Firstly, the boundary line around thefibril was drawn manually with Image Pro Plus to separate the image into afibril structure area and a non-fibril structure area. The grey value of the non-fibril area was substitute with black (0) or white (255) for differentiation (Fig. 12(a)). Then, the totalfibril area and percent of fibril area coverage were calculated. Secondly, the midline for eachfibril was drawn and the total length of thefibril was calculated in the image (Fig. 12(b)). Thirdly, the diameter

2068 1325 832 4151 7808 8295 15929 30891 7549 12571 52111 6495174951 6409 15793 25499 71690 232424 340 268 143 180 409 680 1569 3082 373 553 3303 4900 6200 384 736 1556 5492 12295

Origin_RTFOT4PAV A5 A10 A15 A20 A30 B5 B10 B15 B20 B30 C 5 C10 C15 C20 C30 100 1000 10000 100000 1000000 1E7

N

f

ni

art

S

%

5.

2 @

（

s

el

c

y

c

）

(a)

Origin_RTFOT4PAV A5 A10 A15 A20 A30 B5 B10 B15 B20 B30 C5 C10 C15 C20 C30

100 1000 10000 100000

N

f

ni

art

S

%

5 @

（

s

el

c

y

c

）

(b)

(10)

of eachfibril was drawn, and the average of fibril diameter was calcu-lated (Fig. 12(c)). Finally, the distance between two adjacentfibrils midline was extracted, and the average distance between adjacent fi-brils was calculated.

Assuming that the bitumen microstructure's morphology is isotro-pic, the 2-D ESEM image can be extracted in the 3rd direction. In this way, a simple frame model can be established (Fig. 13). Based on these extracted parameters, the package density (volume fraction) can be calculated: As¼ d2 π 4 ð7Þ Vs¼ 3 As 4 a 2− ffiffiffiffiffi As p 2 þ ffiffiffiffiffi As p 2 3 " # ð8Þ fs¼ Vs a₌₂ ð Þ3 100% ð9Þ

where the Asis the cross-sectional area of afibril; d is the average di-ameter offibril structure; Vsis the volume occupied byfibril structure; a is the average space between two adjacentfibrils, fsthe volume density of thefibril structure.

The ESEM images were segmented and examples can be seen in

Fig. 14. The parameter extraction and quantitative analysis process

were conducted afterward, and the results were illustrated inTable 4. As illustrated inTable 4, thefibril area of the PMB binder increased after RTFOT aging, which agreed with the results of Mikhailenko [48,65]. Only the addition of more than 20% of Rej-B and Rej-C was able to reduce thefibril area significantly. In other cases, the addition of the rejuvenator had no significant effect on the fibril area but mainly

changed the diameter of thefibril structure. The average diameter of fi-bril structure (a) and the average distance between adjacentfibrils (d) increased with a higher content of rejuvenator, especially for B30 and C30. The Asand Vsparameters exhibited similar patterns of change with the addition of a rejuvenator. However, the variation of fsis quite complex, and there is no obvious pattern.

4. Principal component and correlation analysis 4.1. Principal component analysis (PCA)

It is not possible to directly formulate a link between ESEM-induced microstructural changes and the rheology of binders. The PCA was ap-plied to explore the rheological and microstructure data, reduce the di-mensionality of the data and propose a link between various variables. The PCA analysis can enable the identiﬁcation of systematic variations in the data by combining parameters (i.e., rheological and microstruc-ture parameters) that vary in the same (or opposite) manner between different binders and turn these combinations into a new set of param-eters (principal component, PC). Therefore, the PC1 accounts for the maximum amount of variation in the data and PC2 accounts for the maximum of the remaining variation. Usually, theﬁrst two PCs are enough to explain the majority of the data variance (>70%) and gener-ate two graphic outputs called loading plot and score plot [66,67].

In this research, a set of data containing 10 samples and 15 parame-ters is analyzed with PCA. The detailed data can be seen inTable 5. Ac-cording to the results of PCA, theﬁrst two principal components PC1 (69.3%) and PC2 (14.4%) can explain 83.7% of the total variance.

The loading plot inFig. 15(a) provides information about the relation between original variables (i.e., rheology and ESEM structure) and prin-cipals components (PC1 and PC2). The loading plot mainly reﬂects the

(11)

correlation matrix of different parameters. When the vector of the two parameters are close in direction, then it indicates that they are likely to have a positive correlation relationship [66].

FromFig. 15(a), Jnr3.2and PC1 are almost in the same direction,

which indicates that these two variables have a positive correlation

relationship. As Jnr3.2indicates the non-recoverable compliance, it can be deduced that the positive direction of PC1 indicates a binder of low stiffness. In terms of ESEM structural parameters,ﬁbril structure diam-eter (d) and distance between adjacentﬁbrils (a) are negatively corre-lated with the stiffness of the binder. Meanwhile, the formation time

(12)

andfibril length are positively correlated with the stiffness of the binder. PC2 is relatively consistent with the phase angle index (IP), which indi-cates that PC2 can be used to describe the viscoelastic property of binders. However, no ESEM microstructure parameter was found to cor-relate well with the PC2. The percent recovery (R3.2), fatigue life (Nf2.5 and Nf5) and complex modulus index (IM) were influenced by PC1 and PC2, indicating that they were influenced by the binder rheology. Corre-spondingly,fibril area coverage, fibril structure volume (Vs), andfibril package density (fs) are related to the stiffness and the viscoelastic property of the binder.

The results of PCA were also illustrated in a score plot (Fig. 15

(b)), in which each data point corresponds to a binder, and the loca-tion of data points reﬂects the property difference on the scale of PC1 and PC2. According to the loading plot, the negative PC1 indicates binders of high stiffness(low Jnr3.2 and high G*), while the positive PC2 indicates binders of high elasticity (lowδ and high R3.2). As

illustrated inFig. 15(b), the stiffness and elasticity of the binder in-creased during RTFOT aging. The addition of Rej-A mainly recovered viscous property but did not have a signiﬁcant effect on stiffness re-duction. Considering the main component of Rej-A is an aromatic fraction, it indicates that the main role of the aromatics fraction in re-juvenator is to recover viscous property rather than a decrease of stiffness. In contrast, Rej-B is composed of a saturate fraction without any aromatics. The addition of Rej-B dramatically decreased the stiff-ness and increase the elasticity of the binder, suggesting the saturate fraction mainly played a role in reducing the stiffness. The hybrid re-juvenator Rej-C was a mix of aromatics, saturates, polymer and addi-tives, which separates asphaltenes' agglomeration. With the increase of Rej-C content, rejuvenated bitumen's stiffness decreased, and vis-cous property increased. In summary, the composition of the rejuve-nator has a decisive in_{ﬂuence on the rejuvenation mechanism and} the rejuvenation effect it achieves.

(13)

4.2. Correlations between ESEM structural parameters and rheological parameters

If the vectors of two parameters are close in the same (opposite) di-rection, the PCA loading plot indicates a positive (negative) correlation between these two parameters. Based on PCA results inFig. 15, regres-sion analysis was conducted for the ESEM structure parameters and rheological properties. The regression results andﬁtted equation are presented inFig. 16.

FromFig. 16(a),ﬁbril structure diameter (d) and distance between

adjacentfibril (a) exhibited a very significant linear relationship (R2₌ 0.93), indicating that the increase of parameter a often accompanied an increase in parameter d. This finding is coherent with the conclusion in the literature [47,65]. The difference is that their investigation showed a simultaneous decrease of a and d due to aging, while this study exhibited an increase of a and d due to rejuvenation. Meanwhile,

there was a negative linear correlation between formation time and a

(Fig. 16(b)), indicating the addition of rejuvenator made theﬁbril

struc-ture sparse and easy to emerge under the electron beam's illumination.

Fig. 16(c) illustrated an exponential decrease in totalﬁbril length with

an increase in d, and Fig16 (d) demonstrated that Vs decreased linearly with the increase of percent ofﬁbril area coverage. The remarkable cor-relation of ESEM structural parameters indicated that the morphology ofﬁbril structure always obeyed a certain distribution pattern, regard-less of how it changed with aging and rejuvenation.

The ESEM structural parameters are also well correlated with rheological parameters. As illustrated inFig. 16(e) and (f), with the increase of a, Jnr3.2increased linearly while the IMdecreased ex-ponentially. It implies that a sparse ﬁbril structure suggests a binder of low stiffness. From the PCA loading plot,ﬁbril structure density (fs) may have a correlation with percent recovery (R3.2) and fatigue life (Nf5). However,Fig. 16(g) and (h), illustrated that

(14)

there is no correlation between fsand R3.2, and there was a weak ex-ponential correlation between fsand Nf5. It may be the fact that R3.2 and Nf5are inﬂuenced by both stiffness and phase angle of the binder.

5. Discussion

In this paper, the aging and rejuvenation process of binders mani-fests how chemical composition affects their rheological properties

and microstructure. In agreement with previous studies [68–70], it is believed that the change of chemical composition results in the change of colloidal structure, which is subsequently reﬂected in the change in binder microstructure and rheological properties. The colloidal struc-ture can act as a bridge to connect the chemical composition, micro-structure and rheological properties. Evidence to support this statement is as follows.

Firstly, both aging and rejuvenation correspond to microstructural changes in binders as observed in ESEM. In earlier studies [53,62], the increase of asphaltenes in SBS modiﬁed bitumen due to aging was reﬂected by the FTIR and GPC results, which led to an increase of

OB

RTFOT A20 A30 B15 B20 B30 C10 C15 C20 C30

0 20 40 60 80 100 120 140 160 180

132

168

157

138

117

98

42

158

143

133

92 Formation time (s)

Fig. 11. ESEMﬁbril microstructure formation time.

Fig. 12. The process to extract microscopic parameters from ESEM images, (a) definition and calculation of fibril structure area; (b) definition of the midline of fibril structure and calculation of the total length offibril structure; (c) calculation of the fibril structure diameter, (d) calculation of the distance between adjacent fibrils.

Fig. 13. Model for quantification of parameters of fibril structure of the bitumen microstructure (d: averagefibril diameter; a: average space between two adjacent fibrils) [43].

(15)

stiffness and elasticity or an increase of IMand decrease of IPand Jnr3.2. Meanwhile, this phenomenon was also observed in ESEM measure-ments, as theﬁbril structure turned to be denser and harder to be re-vealed (an increase of formation time, area/volume coverage and decrease of a and d). Furthermore, the addition of the rejuvenator in-creased the content of maltenes in bitumen, dein-creased the stiffness and elasticity. The ESEM microstructure is also turned looser and easier to be revealed. For example, when Rej-B's addition increases from 20 to 30%, the IMdecreased from 92.3 to 76.5%. Simultaneously, the average

distance betweenﬁbrils increased, the ﬁbril structure turned to be sparse, and the area coverage decreased from 83.8 to 70.0%.

Secondly, the relationship between rheological properties and ESEM microstructure parameters was revealed with the PCA-loading plot. The inﬂuence of chemical compositional change due to the addition of reju-venators was also illustrated in the PCA-score plot. All the rheological and ESEM structural parameters were explained into two principal components (PC1 and PC2). PC1 explained the stiffness of bitumen, and the parameters with vector direction close to PC1 could also be

Fig. 14. Examples of sample segmentation for aged PMB binder with rejuvenators.

Table 4

Parameters of ESEMﬁbril microstructure.

Binder Fibril area Area coverage Totalﬁbril length d (μm) a (μm) As Vs fs

(μm2

) % (μm) Mean Std. Dev. Mean Std. Dev. (μm2

) (μm3 ) % OB 10,022 89.1 1169.1 9.7 3.0 14.5 4.4 73.6 243.3 63.4 RTFOT 10,376 92.3 1197.4 7.9 1.1 13.0 2.9 48.9 152.4 55.8 A20 9850 87.6 1020.9 10.6 1.6 18.4 6.9 88.8 404.9 51.6 A30 10,122 90.0 852.2 12.6 1.7 20.0 3.9 124.6 584.7 58.9 B15 9915 88.2 1302.3 12.3 1.7 18.6 2.0 119.2 507.0 62.8 B20 9418 83.8 585.7 17.3 2.8 24.2 4.0 236.0 1232.8 69.8 B30 7872 70.0 387.0 26.1 8.3 48.9 17.6 535.7 6719.7 46.0 C10 9843 87.6 1500.2 7.9 2.1 11.2 1.2 49.5 120.0 69.1 C15 9945 88.5 1423.0 9.1 1.7 13.0 2.0 65.5 187.3 67.9 C20 10,046 89.4 1368.0 10.3 1.3 15.9 2.7 83.8 306.9 61.4 C30 9278 82.5 539.6 16.9 3.9 35.3 5.2 225.3 2135.0 38.9

(16)

used to characterize bitumen's stiffness. In this way, the larger totalfibril length and formation time, the smaller a and d, indicating a binder of higher stiffness. PC2 describes the viscoelastic properties of bitumen. Only the phase angle index (IP) showed a significant correlation with PC2. When the viscoelastic property of bitumen is recovered, thefibril structure becomes a“random” state correspondingly. Further investiga-tion is needed to characterize this phenomenon. Furthermore, the PCA-score plot provides information about the influence of chemical compo-sition on microstructure. The addition of aromatics (Rej-A) mainly en-hanced the viscous portion of rheology, and the addition of saturates (Rej\\B) reduces the stiffness.

Thirdly, the quantitative relationship between the rheological prop-erties of binders and microstructure parameters was established by the regression analysis method. For instance, the parameter a (distance be-tween adjacentﬁbril) has shown a signiﬁcant linear positive correlation with Jnr3.2(R2= 0.92) and an exponential negative correlation with IM (R2_{= 0.71), indicating a good connection between ESEM} microstruc-ture and rheology.

In summary, the current observations strengthen the hypothesis that the ESEM microstructure is intimately correlated with bitumen's chemical composition and rheological properties rather than just an irrelevant surface phenomenon. The establishment of the chemo-microstructural-mechanical relationship of bituminous binder

Ta b le 5 Da ta set s fo r p rincipa l co mpo n ent ana lysis . Sample Rheological Parameters ESEM microstructure parameters Nf2.5 Nf5 IM IP R3.2 Jnr3.2 Formation time Fibril Area Area Coverage Total Fibril Length d a As Vs fs cycle cycle % % % kPa-1 s (μ m2) % (μ m) (μ m) (μ m) (μ m2) (μ m3) % Origin 2067.7 340.0 100.0 100.0 40.5 0.03 132 1220.0 89.1 1169.1 9.7 14.5 73.6 243.3 63.4 RTFOT 1325.0 268.0 119.3 79.7 35.0 0.02 168 1809.0 83.9 1197.4 7.9 13.0 48.9 152.4 55.8 A20 15,929.2 1568.9 99.1 93.4 5.4 0.24 157 1392.0 87.6 1020.9 10.6 18.4 88.8 404.9 51.6 A30 30,890.8 3081.9 90.3 97.1 0.0 0.81 138 1120.0 90.0 852.2 12.6 20.0 124.6 584.7 58.9 B15 52,110.7 3303.2 91.2 85.2 2.5 0.37 117 1326.6 88.2 1302.3 12.3 18.6 119.2 507.0 62.8 B20 64,950.8 4900.2 92.3 87.4 2.9 0.44 98 1824.0 83.8 585.7 17.3 24.2 236.0 1232.8 69.8 B30 74,950.8 6200.2 76.5 90.8 0.0 2.37 42 3370.0 70.0 387.0 26.1 48.9 535.7 6719.7 46.0 C10 15,793.3 735.7 109.9 85.3 42.0 0.03 158 1399.0 87.6 1500.2 7.9 11.2 49.5 120.0 69.1 C15 25,498.6 1555.8 101.1 88.4 19.6 0.12 143 1297.5 88.5 1423.0 9.1 13.0 65.5 187.3 67.9 C20 71,690.3 5492.0 91.5 92.4 3.0 0.40 133 1196.0 89.4 1368.0 10.3 15.9 83.8 306.9 61.4 C30 232,424.1 12,295.3 82.0 99.1 0.0 1.23 92 1964.0 82.5 539.6 16.9 35.3 225.3 2135.0 38.9

(17)

10 15 20 25 30 35 40 45 50 5 10 15 20 25 30 10 15 20 25 30 35 40 45 50 40 60 80 100 120 140 160 180 200 400 600 800 1000 1200 1400 1600 5 10 15 20 25 30 70 75 80 85 90 -1000 0 1000 2000 3000 4000 5000 6000 7000 10 15 20 25 30 35 40 45 50 0 1 2 3 10 15 20 25 30 35 40 45 50 75 80 85 90 95 100 105 110 115 120 125 35 40 45 50 55 60 65 70 -5 0 5 10 15 20 25 30 35 40 45 35 40 45 50 55 60 65 70 0 2000 4000 6000 8000 10000 12000 14000 Origin RTFOT A20 A30 B15 B20 B30 C10 C15 C20 C30 Origin RTFOT A20 A30 B15 B20 B30 C10 C15 C20 C30 Origin RTFOT A20 A30 B15 B20 B30 C10 C15 C20 C30 Origin RTFOT A20 A30 B15 B20 B30 C10 C15 C20 C30 Origin RTFOT A20 A30 B15 _B20 B30 C10 C15 C20 C30 Origin RTFOT A20 A30 B15 B20 B30 C10 C15 C20 C30 Origin RTFOT A20 A30 B15 B20 B30 C10 C15 C20 C30 RTFOT Origin A20 A30 B15 B20 B30 C10 C15 C20 C30 d a y = a + b * x d Pearson's r 0.968 Adj. R-Squ 0.930 Intercept 3.00176 ± 0.95 Slope 0.46321 ± 0.03 a y = a + b * x Pearson's -0.92497 Adj. R-Squ 0.83952 Intercept 188.06724 ± 9. Slope -2.96577 ± 0.4 d Tot a l Fibril Lengt h y = a-b*c^x d Reduced C 2.10739 Adj. R-Squa 0.92904 a 9.22982 ± 0.77 b -79.05448 ± 29. c 0.99598 ± 9.20 Vs 3 Area Coverage y = a + b * x Vs Pearson' -0.94459 Adj. R-S 0.88029 Intercept 28590.39751 ± Slope -320.96065 ± 37 Jnr 3.2 (k P a -1 a y = a + b * x Jnr3.2 Pearson's 0.9667 Adj. R-Squ 0.92724 Intercept -0.72223 ± 0.1 Slope 0.06015 ± 0.00 IM a y = a-b*c^x Reduced Ch 41.12593 Adj. R-Squar 0.71659 a 77.46972 ± 6.86 b -97.90117 ± 50. c 0.90974 ± 0.045 R3.2 fs y = a + b * x R3.2 Pearson' 0.42036 Adj. R-S 0.08523 Intercept -29.53633 ± Slope 0.73674 ± 0.5 Nf5 fs y = a-b*c^x N_f5 Reduced Ch 4339102.71409 Adj. R-Squar 0.65626 a 2278.0714 ± 891. b -1.77063E7 ± 8.09 c 0.82543 ± 0.0971

(18)

improves the understanding of the underlying process and mechanism of aging and rejuvenation. This will enable the proper design formula for the rejuvenator or develop an anti-aging agent.

6. Conclusions

In this research, original, aged and rejuvenated SBS modified binders were investigated to establish their chemical, microstructural, and me-chanical relationship to reveal the underlying rejuvenation mechanism. Three types of rejuvenators were used to rejuvenate the aged SBS mod-ified bitumen at five different dosages. The rheological properties of binders were measured with DSR. The microstructure of rejuvenated binders was observed with ESEM, and a series of microstructure param-eters were extracted. The rheological and microstructure characteristics were analyzed with PCA and regression analysis, and the conclusions are the following:

• Chemical compositional changes of binders correspond with the rhe-ological and microstructure alterations. The saturate fraction increase mainly decreases the binder stiffness, and the corresponding micro-structure becomes sparse. Moreover, the addition of aromatic fraction mainly increases the viscous portion, and the corresponding micro-structure becomes more random rather than perpendicular. Further addition of the SBS polymer and rejuvenator led to an increase in the viscous portion and made theﬁbril microstructure visible. • From PCA results, microstructure and rheological parameters can be

combined into two principal components (PC1 and PC2), in which PC1 accounts for the stiffness and PC2 accounts for the viscoelastic property of binders. Among the microstructure parameters obtained from ESEM,fibril structure formation time, total fibril length, fibril di-ameter and distance between adjacentfibril structure explain the stiffness of binder, and volume density offibril structure is associated with both stiffness and viscoelastic property.

• According to the regression analysis, the distance between adjacent ﬁ-brils exhibited a signiﬁcant correlation with Jnr3.2(R2= 0.92) and complex modulus index (R2_{= 0.71), and the volume density of}

ﬁbril structure showed a correlation with fatigue life (R2 _{= 0.66). It} strengthens the hypothesis that the microstructure is intimately asso-ciated with the binder composition and rheological properties rather than just an irrelevant surface phenomenon.

Credit author statement

The authors conﬁrm contribution to the paper as follows: study con-ception and design: Xueyan Liu, Peng Lin, Panos Apostolidis, Sandra Erkens; data collection: Peng Lin and Yi Zhang; analysis and interpreta-tion of results: Xueyan Liu, Peng Lin and Panos Apostolidis; draft manu-script preparation: Peng Lin, Xueyan Liu, Panos Apostolidis and Shisong Ren. All authors reviewed the results and approved theﬁnal version of the manuscript.

Declaration of Competing Interest

The authors declare that they have no known competingﬁnancial interests or personal relationships that could have appeared to in ﬂu-ence the work reported in this paper.

Acknowledgments

Special thanks to Laurent Porot (Kraton Polymers BV) for supplying the SBS copolymer and the SBS modiﬁed bitumen used in this research. We would like to thank Arjan Thijssen from Microlab Civil Engineering & Geosciences at the Delft University of Technology for ESEM technique and analysis support.

References

[1] Asphalt Institute, European Bitumen Association, The Bitumen Industry: a Global Perspective: Production, Chemistry Use, Speciﬁcation and Occupational Exposure, 2015.

[2] F. Chaignon, P. Eapa, Asphalt Fig. 2018 (2018) 13.

[3] M. Zaumanis, R.B. Mallick, Review of very high-content reclaimed asphalt use in plant-produced pavements: state of the art, Int. J. Pav. Eng. 16 (2015) 39–55, https://doi.org/10.1080/10298436.2014.893331.

[4] N. Tran, A. Taylor, P. Turner, C. Holmes, L. Porot, Effect of rejuvenator on perfor-mance characteristics of high RAP mixture, Road Mater. Pav. Design. 18 (2017) 183–208,https://doi.org/10.1080/14680629.2016.1266757.

[5] R. Tauste, F. Moreno-Navarro, M. Sol-Sánchez, M.C. Rubio-Gámez, Understanding the bitumen ageing phenomenon: a review, Constr. Build. Mater. 192 (2018) 593–609,https://doi.org/10.1016/j.conbuildmat.2018.10.169.

[6] J.C. Petersen, Transportation Research Circular E-C140: A Review of the Fundamen-tals of Asphalt Oxidation: Chemical, Physicochemical, Physical Property, and Dura-bility Relationships, Transportation Research Board of the National Academies, Washington, DC, 2009.

[7] F. Zhou, P. Karki, Rejuvenator Laboratory Characterization and Field Performance, 2021 218.

[8] W.S. Mogawer, A. Austerman, R. Roque, S. Underwood, L. Mohammad, J. Zou, Ageing and rejuvenators: evaluating their impact on high RAP mixtures fatigue cracking characteristics using advanced mechanistic models and testing methods, Road Mater. Pav. Design. 16 (2015) 1–28,https://doi.org/10.1080/14680629.2015. 1076996.

[9] L. Porot, D. Broere, M. Wistuba, J. Grönniger, Asphalt and binder evaluation of as-phalt mix with 70% reclaimed asas-phalt, Road Mater. Pav. Design. 18 (2017) 66–75, https://doi.org/10.1080/14680629.2017.1304259.

[10] J. Shen, S. Amirkhanian, J. Aune Miller, Effects of rejuvenating agents on superpave mixtures containing reclaimed asphalt pavement, J. Mater. Civ. Eng. 19 (2007) 376–384,https://doi.org/10.1061/(ASCE)0899-1561(2007)19:5(376).

[11] J. Shen, S. Amirkhanian, S.-J. Lee, The effects of rejuvenating agents on recycled aged CRM binders, Int. J. Pav. Eng. 6 (2005) 273–279, https://doi.org/10.1080/ 10298430500439319.

[12] K.-A.N. Johnson, S.A.M. Hesp, Effect of waste engine oil residue on quality and dura-bility of SHRP materials reference library binders, Transp. Res. Rec. 2444 (2014) 102–109,https://doi.org/10.3141/2444-12.

[13] P. Karki, F. Zhou, Effect of rejuvenators on rheological, chemical, and aging proper-ties of asphalt binders containing recycled binders, Transp. Res. Rec. 2574 (2016) 74–82,https://doi.org/10.3141/2574-08.

[14] T. Baghaee Moghaddam, H. Baaj, The use of rejuvenating agents in production of recycled hot mix asphalt: a systematic review, Constr. Build. Mater. 114 (2016) 805–816,https://doi.org/10.1016/j.conbuildmat.2016.04.015.

[15] X. Yu, M. Zaumanis, S. dos Santos, L.D. Poulikakos, Rheological, microscopic, and chemical characterization of the rejuvenating effect on asphalt binders, Fuel. 135 (2014) 162–171,https://doi.org/10.1016/j.fuel.2014.06.038.

[16] D. Ghosh, M. Turos, E. Johnson, M. Marasteanu, Rheological characterization of as-phalt binders treated with bio sealants for pavement preservation, Can. J. Civ. Eng. 45 (2018) 407–412,https://doi.org/10.1139/cjce-2017-0058.

[17] M.N. Siddiqui, M.F. Ali, Studies on the aging behavior of the Arabian asphalts, Fuel. 78 (1999) 1005–1015,https://doi.org/10.1016/S0016-2361(99)00018-6. [18] J. Wang, T. Wang, X. Hou, F. Xiao, Modelling of rheological and chemical properties

of asphalt binder considering SARA fraction, Fuel. 238 (2019) 320–330,https://doi. org/10.1016/j.fuel.2018.10.126.

[19] A. Hung, E.H. Fini, Surface morphology and chemical mapping of UV-aged thinﬁlms of bitumen, ACS Sustain. Chem. Eng. 8 (2020) 11764–11771,https://doi.org/10. 1021/acssuschemeng.0c03877.

[20] A.A. Hussein, R.P. Jaya, N. Abdul Hassan, H. Yaacob, G.F. Huseien, M.H.W. Ibrahim, Performance of nanoceramic powder on the chemical and physical properties of bi-tumen, Constr. Build. Mater. 156 (2017) 496–505, https://doi.org/10.1016/j. conbuildmat.2017.09.014.

[21] K. Yan, H. Zhang, H. Xu, Effect of polyphosphoric acid on physical properties, chem-ical composition and morphology of bitumen, Constr. Build. Mater. 47 (2013) 92–98,https://doi.org/10.1016/j.conbuildmat.2013.05.004.

[22] Z.-G. Feng, J.-Y. Yu, H.-L. Zhang, D.-L. Kuang, L.-H. Xue, Effect of ultraviolet aging on rheology, chemistry and morphology of ultraviolet absorber modiﬁed bitumen, Mater. Struct. 46 (2013) 1123–1132.

[23] H.L. Zhang, J.Y. Yu, Z.G. Feng, L.H. Xue, S.P. Wu, Effect of aging on the morphology of bitumen by atomic force microscopy, J. Microsc. 246 (2012) 11–19,https://doi.org/ 10.1111/j.1365-2818.2011.03578.x.

[24] X. Lu, M. Langton, P. Olofsson, P. Redelius, Wax morphology in bitumen, J. Mater. Sci. 40 (2005) 1893–1900.

[25]S. Bearsley, A. Forbes, R.G. Haverkamp, Direct observation of the asphaltene struc-ture in paving-grade bitumen using confocal laser-scanning microscopy, J. Microsc. 215 (2004) 149–155.

[26] D. Sun, G. Sun, X. Zhu, Q. Pang, F. Yu, T. Lin, Identiﬁcation of wetting and molecular diffusion stages during self-healing process of asphalt binder viaﬂuorescence mi-croscope, Constr. Build. Mater. 132 (2017) 230–239,https://doi.org/10.1016/j. conbuildmat.2016.11.137.

[27] D. Lesueur, The colloidal structure of bitumen: consequences on the rheology and on the mechanisms of bitumen modiﬁcation, Adv. Colloid Interf. Sci. 145 (2009) 42–82,https://doi.org/10.1016/j.cis.2008.08.011.

[28] F. Handle, J. Füssl, S. Neudl, D. Grossegger, L. Eberhardsteiner, B. Hofko, M. Hospodka, R. Blab, H. Grothe, The bitumen microstructure: aﬂuorescent approach, Mater. Struct. 49 (2016) 167–180.

(19)

[29] R. Jahangir, D. Little, A. Bhasin, Evolution of asphalt binder microstructure due to tensile loading determined using AFM and image analysis techniques, Int. J. Pav. Eng. 16 (2015) 337–349.

[30] A. Jäger, R. Lackner, C. Eisenmenger-Sittner, R. Blab, Identiﬁcation of microstructural components of bitumen by means of atomic force microscopy (AFM), PAMM: Pro-ceedings in Applied Mathematics and Mechanics, Wiley Online Library 2004, pp. 400–401.

[31] H. Soenen, J. Besamusca, H.R. Fischer, L.D. Poulikakos, J.-P. Planche, P.K. Das, N. Kringos, J.R.A. Grenfell, X. Lu, E. Chailleux, Laboratory investigation of bitumen based on round robin DSC and AFM tests, Mater. Struct. 47 (2014) 1205–1220, https://doi.org/10.1617/s11527-013-0123-4.

[32] J. Stulirova, K. Pospisil, Observation of bitumen microstructure changes using scan-ning electron microscopy, Road Mater. Pav. Design. 9 (2008) 745–754,https://doi. org/10.1080/14680629.2008.9690148.

[33] K. Pospíšil, A. Frỳbort, A. Kratochvíl, J. Macháčková, Scanning electron microscopy method as a tool for the evaluation of selected materials microstructure, Transact. Trans. Sci. 1 (2008) 13–20.

[34] M. Mazumder, R. Ahmed, A. Wajahat Ali, S.-J. Lee, SEM and ESEM techniques used for analysis of asphalt binder and mixture: a state of the art review, Constr. Build. Mater. 186 (2018) 313–329,https://doi.org/10.1016/j.conbuildmat.2018.07.126. [35] L. Loeber, O. Sutton, J. Morel, J.-M. Valleton, G. Muller, New direct observations of

as-phalts and asphalt binders by scanning electron microscopy and atomic force mi-croscopy, J. Microsc. 182 (1996) 32–39,https://doi.org/10.1046/j.1365-2818.1996. 134416.x.

[36] J.-B. Donnet, J. Ducret, M. Kennel, E. Papirer, Electron microscopic observations of the morphology of bitumens, Fuel. 56 (1977) 97–100.

[37] G.D. Daniatos, Design and construction of an atmospheric or environmental SEM (part 1), Scanning. 4 (1981) 9–20.

[38] G.D. Danilatos, R. Postle, Design and construction of an atmospheric or environmen-tal SEM—2, Micron 14 (1983) (1969) 41–52.

[39] G.D. Danilatos, Design and construction of an atmospheric or environmental SEM (part 3), Scanning. 7 (1985) 26–42.

[40] A.S. Bhurke, E.E. Shin, L.T. Drzal, Fracture morphology and fracture toughness mea-surement of polymer-modiﬁed asphalt concrete, Transp. Res. Rec. 1590 (1997) 23–33,https://doi.org/10.3141/1590-04.

[41] S.J. Rozeveld, E.E. Shin, A. Bhurke, L. France, L.T. Drzal, Network morphology of straight and polymer modiﬁed asphalt cements, Microsc. Res. Tech. 38 (1997) 529–543, https://doi.org/10.1002/(SICI)1097-0029(19970901)38:5<529::AID-JEMT11>3.0.CO;2-O.

[42] M.C. Hawley, L.T. Drzal, G. Baladi, Y.-J. Lee, Final Report on Polymers in Bituminous Mixtures - PHASE II,https://trid.trb.org/view/4965721997. (Accessed 25 June 2020).

[43] K. Stangl, A. Jäger, R. Lackner, Microstructure-based identiﬁcation of bitumen perfor-mance, Road Mater. Pav. Design. 7 (2006) 111–142,https://doi.org/10.1080/ 14680629.2006.9690061.

[44] L.C. Michon, T.M. Williams, F.P. Miknis, J.-P. Planche, D. Martin, Use of the environ-mental scanning electron microscope to investigate three polymer modiﬁedasphalts, Pet. Sci. Technol. 16 (1998) 797–809,https://doi.org/10.1080/10916469808949812. [45] J. Gaskin, On Bitumen Microstructure and the Effects of Crack Healing, (Notingham

University PhD Thesis) 265.

[46] P. Mikhailenko, H. Kadhim, H. Baaj, Observation of bitumen microstructure oxida-tion and blending with ESEM, Road Mater. Pav. Design. 18 (2017) 216–225, https://doi.org/10.1080/14680629.2017.1304251.

[47] P. Mikhailenko, C. Kou, H. Baaj, S. Tighe, Observation of Polymer Modiﬁed Asphalt Microstructure by ESEM, 2017 11.

[48] P. Mikhailenko, C. Kou, H. Baaj, L. Poulikakos, A. Cannone-Falchetto, J. Besamusca, B. Hofko, Comparison of ESEM and physical properties of virgin and laboratory aged asphalt binders, Fuel. 235 (2019) 627–638,https://doi.org/10.1016/j.fuel.2018.08. 052.

[49] J. Zhu, B. Birgisson, N. Kringos, Polymer modiﬁcation of bitumen: advances and chal-lenges, Eur. Polym. J. 54 (2014) 18–38.

[50] J.C. Munera, E.A. Ossa, Polymer modiﬁed bitumen: optimization and selection, Mater. Des. 62 (2014) (1980-2015) 91–97,https://doi.org/10.1016/j.matdes.2014. 05.009.

[51] G. Polacco, J. Stastna, D. Biondi, L. Zanzotto, Relation between polymer architecture and nonlinear viscoelastic behavior of modiﬁed asphalts, Curr. Opin. Colloid Inter-face Sci. 11 (2006) 230–245,https://doi.org/10.1016/j.cocis.2006.09.001. [52] C. Yan, W. Huang, P. Lin, Y. Zhang, Q. Lv, Chemical and rheological evaluation of

aging properties of high content SBS polymer modiﬁed asphalt, Fuel. 252 (2019) 417–426.

[53] P. Lin, C. Yan, W. Huang, Y. Li, L. Zhou, N. Tang, F. Xiao, Y. Zhang, Q. Lv, Rheological, chemical and aging characteristics of high content polymer modiﬁed asphalt, Constr. Build. Mater. 207 (2019) 616–629,https://doi.org/10.1016/j.conbuildmat. 2019.02.086.

[54] N.I. Md, F.M. Jakarni Yusoff, V.H. Nguyen, M.R. Hainin, G.D. Airey, Modelling the rhe-ological properties of bituminous binders using mathematical equations, Constr. Build. Mater. 40 (2013) 174–188,https://doi.org/10.1016/j.conbuildmat.2012.09. 105.

[55] H.C. Booij, G.P.J.M. Thoone, Generalization of Kramers-Kronig transforms and some approximations of relations between viscoelastic quantities, Rheol. Acta 21 (1982) 15–24,https://doi.org/10.1007/BF01520701.

[56] M. Oshone, E. Dave, J.S. Daniel, G.M. Rowe, Prediction of phase angles from dynamic modulus data and implications for cracking performance evaluation, Road Mater. Pav. Design. 18 (2017) 491–513,https://doi.org/10.1080/14680629.2017.1389086. [57] M.C. Cavalli, M. Zaumanis, E. Mazza, M.N. Partl, L.D. Poulikakos, Effect of ageing on

the mechanical and chemical properties of binder from RAP treated with bio-based rejuvenators, Compos. Part B 141 (2018) 174–181,https://doi.org/10.1016/ j.compositesb.2017.12.060.

[58] C. Hintz, R. Velasquez, C. Johnson, H. Bahia, Modiﬁcation and validation of linear am-plitude sweep test for binder fatigue speciﬁcation, Transp. Res. Rec. 2207 (2011) 99–106,https://doi.org/10.3141/2207-13.

[59] N. Saboo, R. Kumar, P. Kumar, A. Gupta, Ranking the rheological response of SBS-and EVA-modified bitumen using MSCR SBS-and LAS tests, J. Mater. Civ. Eng. 30 (2018), 04018165,https://doi.org/10.1061/(ASCE)MT.1943-5533.0002367. [60] X. Lu, U. Isacsson, Influence of styrene-butadiene-styrene polymer modification on

bitumen viscosity, Fuel. 76 (1997) 1353–1359, https://doi.org/10.1016/S0016-2361(97)00144-0.

[61] G.D. Airey, Rheological properties of styrene butadiene styrene polymer modiﬁed road bitumens, Fuel. 82 (2003) 1709–1719,https://doi.org/10.1016/S0016-2361 (03)00146-7.

[62] Peng Lin, Huang Weidong, Liu Xueyan, Apostolids Panos, Haopeng Wang, Yan Chuanqi, Laboratory evaluation of the effects of long-term aging on high-content polymer-modiﬁed asphalt binder, J. Mater. Civ. Eng. 32 (2020), 04020157, https://doi.org/10.1061/(ASCE)MT.1943-5533.0003208.

[63] M. Dirand, V. Chevallier, E. Provost, M. Bouroukba, D. Petitjean, Multicomponent parafﬁn waxes and petroleum solid deposits: structural and thermodynamic state, Fuel. 77 (1998) 1253–1260,https://doi.org/10.1016/S0016-2361(98)00032-5. [64] X. Lu, P. Sjövall, H. Soenen, M. Andersson, Microstructures of bitumen observed by

environmental scanning electron microscopy (ESEM) and chemical analysis using time-of-ﬂight secondary ion mass spectrometry (TOF-SIMS), Fuel. 229 (2018) 198–208,https://doi.org/10.1016/j.fuel.2018.05.036.

[65] P. Mikhailenko, H. Kadhim, H. Baaj, Observation of bitumen microstructure oxida-tion and blending with ESEM, Road Mater. Pav. Design. 18 (2017) 216–225, https://doi.org/10.1080/14680629.2017.1304251.

[66] I.T. Jolliffe, Principal Component Analysis, Springer New York, NY, 2002. [67] A. Margaritis, H. Soenen, E. Fransen, G. Pipintakos, G. Jacobs, J. Blom, W. Van Den

Bergh, Identiﬁcation of ageing state clusters of reclaimed asphalt binders using prin-cipal component analysis (PCA) and hierarchical cluster analysis (HCA) based on chemo-rheological parameters, Constr. Build. Mater. 244 (2020) 118276,https:// doi.org/10.1016/j.conbuildmat.2020.118276.

[68] W.J. Halstead, Relation of Asphalt Chemistry to Physical Properties and Speciﬁca-tions, Virginia Transportation Research Council, 1984.

[69] X. Yu, N.A. Burnham, S. Granados-Focil, M. Tao, Bitumen’s microstructures are corre-lated with its bulk thermal and rheological properties, Fuel. 254 (2019) 115509, https://doi.org/10.1016/j.fuel.2019.05.092.

[70] P. Redelius, H. Soenen, Relation between bitumen chemistry and performance, Fuel. 140 (2015) 34–43.